Autism Prevalence Dashboard

Summary

This report brings together three public CDC datasets on autism spectrum disorder into a single interactive Power BI dashboard. It lets a viewer explore how reported prevalence among U.S. children has changed from 2000 to 2022, and how it differs by data source, state, sex, and race — all through one page of connected visuals and slicers.

Context & Problem

Autism prevalence figures are widely cited but rarely easy to explore. The numbers come from several different surveillance programs, span two decades, and vary considerably from state to state and group to group. The goal of this project was to turn that scattered public data into a clear, interactive view where anyone can answer their own questions: Is prevalence rising? How much does it depend on which data source you look at? Where is it highest? Do the patterns differ for boys and girls, or across racial groups?

Objectives

• Track how autism prevalence has changed over time (2000–2022).
• Compare estimates across the different CDC data sources.
• Show geographic differences across U.S. states on a map.
• Compare prevalence by sex (boys vs. girls) and by race.
• Let users filter and drill into the figures that matter to them.

Data Source

Three public datasets published by the CDC's autism data & research program (cdc.gov/autism/data-research), loaded from CSV:

• Prevalence Estimates Over Time by State — prevalence per 1,000 children by state and year.
• ADDM Estimates by Race — prevalence broken down by racial/ethnic group.
• ADDM Estimates by Sex — prevalence for boys vs. girls.

Together they cover U.S. states from 2000 to 2022, with prevalence expressed per 1,000 children. The prevalence figures are drawn from several surveillance sources — the ADDM Network, Medicaid, the National Survey of Children's Health, and Special Education Child Count — each reported with lower and upper confidence intervals.

Approach

• ETL in Power Query: loaded the three CSV files, set correct data types, and handled missing values (blank confidence intervals and NA entries in the race table) so they didn't distort the visuals.
• Data modeling: kept the prevalence, sex, and race data as separate tables and related them on their shared Year and State fields where the grain allowed — which is what determines how far each slicer reaches across the report.
• DAX: built calculated dimension tables for the model — including a unified State dimension drawn from all three source tables — plus measures that return the highest- and lowest-prevalence states and their values to drive the KPI cards at the top of the page.
• Report design: built a single, focused page where the visuals cross-filter and the slicers are scoped to the questions they apply to.

Sample DAX

StateDim =
VAR States =
    DISTINCT(
        UNION(
            SELECTCOLUMNS(
                'ADDM-Estimates-By-Race',
                "State", 'ADDM-Estimates-By-Race'[State]
            ),
            SELECTCOLUMNS(
                'ADDM-Estimates-By-Sex',
                "State", 'ADDM-Estimates-By-Sex'[State]
            ),

            SELECTCOLUMNS(
                'Prevalence-Estimates-Over-Time-by-State',
                "State", 'Prevalence-Estimates-Over-Time-by-State'[State]
            )
        )
    )
RETURN
    ADDCOLUMNS(
        States,
        "StateSort",
        IF(
            [State] = "Total",
            0,
            RANKX(
                FILTER(States, [State] <> "Total"),
                [State],
                ,
                ASC
            )
        )
    )

StateHighestPreval =
    VAR StateTbl = (
        ADDCOLUMNS(
            VALUES('Prevalence-Estimates-Over-Time-by-State'[State]),
            "TotalPrev", CALCULATE([TotalPreval])
        )
    )
   VAR  TopState =(
    TOPN(1,
    StateTbl,
    [TotalPrev], DESC,
    'Prevalence-Estimates-Over-Time-by-State'[State], ASC
    )

   )
   RETURN
   CONCATENATEX(TopState,
   'Prevalence-Estimates-Over-Time-by-State'[State],
   ", ")

What's on the Dashboard

A single page with two KPI cards, four visuals, and three slicers:

• KPI cards — the state with the highest prevalence and the state with the lowest, each with its value.
• Line chart — Prevalence Over Time: prevalence per 1,000 children by year, split by data source (Medicaid, Special Education Child Count, and others).
• Clustered column — Boys vs. Girls: prevalence by sex across years.
• Filled map — Prevalence by State: a geographic view of prevalence across the U.S.
• Clustered column — Race Comparison: prevalence for White, Black, and Hispanic children over time, highlighting disparities.

Interactivity: a Data Set slicer filters every visual except sex and race; a Year slicer drives the map and the two KPI cards; and a State slicer applies to the line, sex, and race charts.

Key Insights

• Prevalence has risen smoothly over the years. In the ADDM Network data, autism prevalence has increased steadily since 2000, reaching a local maximum around 2010 before continuing its upward trend.
• Boys are affected far more than girls. Prevalence among boys runs roughly 3× that of girls on average across the years measured.
• Racial groups are diverging in their trajectories. Prevalence among White children has grown smoothly across the period. Since 2018, however, Black and Hispanic children have seen the steepest rate of increase.
• A handful of states stand out in 2022. California, New Mexico, Minnesota, Missouri, Tennessee, Pennsylvania, and Georgia report the highest estimates, in the 30+ per 1,000 range.

Note: these figures vary considerably by data source and year, so the slicers should be applied with care — comparisons are only meaningful within a single data source and time frame.